General Slowness 2990wx build


Recommended Posts

Included all diagnostics

Quote

 

Threadripper 2990wx

ASRock - X399 Phantom Gaming 6

8x 16gb (128gb) Unbuffered ECC (Crucial)

8x 12tb Ironwolf HDD's

1x 512gb 960 Evo NVMe

1x 1tb Sabrent NVMe

1x RTX2060

 

 

I have gone into the Mobo and disabled c-states, which let my server stay up longer than 15 minutes

I have gone through all VM's and set them to CPU pin on the correct numa node

I have done CPU Isolation

I did do a short test on the RAM with no errors (all new) and i reseated

I did reseat, with new thermal paste, the CPU

 

There is a general lag/latency/slowness everywhere, not just with VM's. Even my plex will end up just hanging up/locking up when running optimize database.

I'm pretty sure I did the following but i'll have to check when i get home to be sure i did.

Quote

/usr/local/sbin/zenstates --c6-disable
/usr/local/sbin/emhttp &

 

I still need to update my BIOS on the mobo, i'm not sure how out of date it is (just got the board)

I dont know what else to try, what else to look for. I'm at a loss, and at a tipping point of frustration.

 

*Note* Syslog does show the following but from what i read i should just ignore it?

 

Quote

Jan  6 10:51:58 RemServer root: Fix Common Problems: Error: Machine Check Events detected on your server
Jan  6 10:51:58 RemServer root: mcelog: ERROR: AMD Processor family 23: mcelog does not support this processor.  Please use the edac_mce_amd module instead.
Jan  6 10:51:58 RemServer root: CPU is unsupported
Jan  6 10:52:00 RemServer root: Fix Common Problems: Error: Machine Check Events detected on your server
Jan  6 10:52:00 RemServer root: mcelog: ERROR: AMD Processor family 23: mcelog does not support this processor.  Please use the edac_mce_amd module instead.
Jan  6 10:52:00 RemServer root: CPU is unsupported

 

*EDIT* I noticed today, even though i have plex set to use any cores it wants its only using 1 thread, capping it out at 100% during streams, wont use any other cores. And obviously keeps having issues because of it.

*EDIT 2* After some troubleshooting it looks like something with my Pinning was maybe corrupted. I reset the pinning and redid it all and now plex is using multiple cores when trans-coding, however its still using 1 core when loading menu's or doing optimization and caps that core to 100%. I've not solved any of my other issues yet. I did update my BIOS with no noticeable change to anything. Plex menu loads forever, often crash's, hard linking through RADARR or SONARR takes forever, VM's are generally sluggish like high latency,

 

 

remserver-diagnostics-20200106-1052.zip

Edited by Remamian
*EDIT2*
Link to comment

First and foremost, do NOT disable Global C State control. With the latest BIOS, Global C State Control is no longer required to be disabled. It, in fact, improves performance when enabled on my 2990WX.

I actually don't remember ever needing to disable it for stability - I believe it's a Ryzen problem and not a TR4 thing.

 

Next, have you checked your CPU frequency while running? Is it thermal throttling? Are you using water cooling?

I have seen several recent posts at various places about people complaining about poor performance on TR4, which turned out to be gunked up water cooling pump (especially the AIO kind).

 

Lastly, docker constantly and inexplicably loading 100% on a single core is a symptom of pinning an isolated core to the docker. Isolation = core can ONLY be used by VM.

Putting ANY isolated core on a docker will eventually cause that core to be loaded to 100% as the docker gets into a loop of trying to use a forbidden core.

Edited by testdasi
  • Like 1
Link to comment
3 hours ago, testdasi said:

First and foremost, do NOT disable Global C State control. With the latest BIOS, Global C State Control is no longer required to be disabled. It, in fact, improves performance when enabled on my 2990WX.

I actually don't remember ever needing to disable it for stability - I believe it's a Ryzen problem and not a TR4 thing.

 

Next, have you checked your CPU frequency while running? Is it thermal throttling? Are you using water cooling?

I have seen several recent posts at various places about people complaining about poor performance on TR4, which turned out to be gunked up water cooling pump (especially the AIO kind).

 

Lastly, docker constantly and inexplicably loading 100% on a single core is a symptom of pinning an isolated core to the docker. Isolation = core can ONLY be used by VM.

Putting ANY isolated core on a docker will eventually cause that core to be loaded to 100% as the docker gets into a loop of trying to use a forbidden core.

 

I am using the Enermax Cooler, but i've done some hard testing with no thermal throttle. Right now its OC'd to 3.3ghz. It gets to about 55c when under 80%+ load. I'm thinking about going up to 3.8ghz OC. I did check pre 3.3ghz OC and during full load (after 25 minutes) there was no thermal throttle, i was having this issue back then as well.

 

I will state that the VM's run great, just there is a noticeable 'lag' to them but once they get going on whatever it was it runs fine.

 

When i had C-States enabled (Albeit was 6.7.7 or whatever) the server would stay on for 15 minutes then shut down. Disabling C-States allowed it to perpetually stay on.

 

As far as pinning is concerned, here's my layout. The isolated cores are Numa Nodes 1/2, nothing but VM's are pinned to those nodes. 

 

image.thumb.png.f8f627aba4cb7f59be1092d75ed403e0.png

Link to comment
On 1/6/2020 at 5:10 PM, Remamian said:

There is a general lag/latency/slowness everywhere, not just with VM's. Even my plex will end up just hanging up/locking up when running optimize database

I'm no expert but i've read that 6.8.0 and higher with plex installations are struggling a lot due to the incredible bad smb performance introduced with 6.8.0.

If possible and not to much trouble i'd try downgrading to 6.7.2 and see if the "sluggishness" remains, because for me 6.8.0 was nearly unusuable and i have many services depending on smb and 6.7.2 handles it just fine.

Edited by je82
Link to comment

Also, try putting everything back to stock clock and see what happens. I've seen reports where overclocking either memory or cpu can cause massive slowdowns. BTW, make sure your memory is running at cpu / motherboard approved frequencies, just because the memory is rated for a certain speed doesn't mean the motherboard / cpu can actually drive it to that speed reliably.

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.